**4. Conclusions**

Results show that near infrared spectroscopy combined with multivariate calibration can be useful as a predictive technique for olive tree pruning biomass analysis, despite its specific compositional features characterized by its high content and variability of the extractives fraction. The predictability of the models in this study is, in principle, limited for the low number of available samples and low compositional variability of calibration set. In spite of this, the biomass components analyzed and predicted with the best model exhibit R2cal values of 0.9 or near of 0.9, and acceptable values of prediction errors. The exception was total lignin, which exhibit a poor linear fit (R2cal < 0.6) and greater prediction errors. Thus, further work is needed in order to improve its potential as a prediction tool and this implies that to increase the number and compositional variability of the sample set, both calibration and external validation are needed.

**Author Contributions:** Conceptualization, M.J.N. and E.C.; methodology, M.J.N., J.L.F.; validation, J.L.F.; formal analysis, F.S.; investigation, M.J.N., M.B., F.S., J.L.F.; writing—original draft preparation, M.J.N., P.M.; writing—review and editing, P.M., M.J.N.

**Funding:** This research was partially funded by Ministerio de Economía y Competitividad (MINECO) (Spain) Reference ENE2011-29112-C02-01 and ENE2011-29112-C02-02 including FEDER funds and EN2017-85819-C2-2-R (AEI/UE).

**Acknowledgments:** Authors would like to thank to MINECO (Spain) Reference ENE2011-29112 including FEDER funds.

**Conflicts of Interest:** The authors declare no conflict of interest.
